MORPH
SIGN IN SIGN UP
kanaka / mal UNCLAIMED

mal - Make a Lisp

0 0 0 Assembly
Normal View History Raw
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
--\i init.sql
@io.sql
plsql: refactor with memory pool. step5,7,8 basics. Ran into the same problem with macroexpand that prevented TCO in step5. Apparently objects as parameters cannot be re-assigning (or copied and re-assigned even) in a loop. When trying to do so, the server process crashes with a memory error. Move away from try to allocate and pass full structures around. Instead pass integer references into a memory pool of the objects (nested table of mal objects) and then pass this memory pool around by reference. Passing the memory pool around all over the place is ugly, but it allows TCO and macroexpand to work at all and seems to have better performance anyways.
2016-04-07 01:22:15 -05:00
CREATE OR REPLACE PACKAGE mal IS
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
plsql: stepA basics. Fix step6 argument processing.
2016-04-29 00:11:59 -05:00
FUNCTION MAIN(args varchar DEFAULT '()') RETURN integer;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
plsql: refactor with memory pool. step5,7,8 basics. Ran into the same problem with macroexpand that prevented TCO in step5. Apparently objects as parameters cannot be re-assigning (or copied and re-assigned even) in a loop. When trying to do so, the server process crashes with a memory error. Move away from try to allocate and pass full structures around. Instead pass integer references into a memory pool of the objects (nested table of mal objects) and then pass this memory pool around by reference. Passing the memory pool around all over the place is ugly, but it allows TCO and macroexpand to work at all and seems to have better performance anyways.
2016-04-07 01:22:15 -05:00
END mal;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
/
plsql: refactor with memory pool. step5,7,8 basics. Ran into the same problem with macroexpand that prevented TCO in step5. Apparently objects as parameters cannot be re-assigning (or copied and re-assigned even) in a loop. When trying to do so, the server process crashes with a memory error. Move away from try to allocate and pass full structures around. Instead pass integer references into a memory pool of the objects (nested table of mal objects) and then pass this memory pool around by reference. Passing the memory pool around all over the place is ugly, but it allows TCO and macroexpand to work at all and seems to have better performance anyways.
2016-04-07 01:22:15 -05:00
CREATE OR REPLACE PACKAGE BODY mal IS
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
plsql: stepA basics. Fix step6 argument processing.
2016-04-29 00:11:59 -05:00
FUNCTION MAIN(args varchar DEFAULT '()') RETURN integer IS
plsql: self-hosting. Long strings, refactor io.
2016-05-03 00:28:33 -05:00
line CLOB;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
plsql: step2,3 basics.
2016-04-04 23:37:52 -05:00
-- read
FUNCTION READ(line varchar) RETURN varchar IS
BEGIN
RETURN line;
END;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
plsql: step2,3 basics.
2016-04-04 23:37:52 -05:00
-- eval
FUNCTION EVAL(ast varchar, env varchar) RETURN varchar IS
BEGIN
RETURN ast;
END;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
plsql: step2,3 basics.
2016-04-04 23:37:52 -05:00
-- print
FUNCTION PRINT(exp varchar) RETURN varchar IS
BEGIN
RETURN exp;
END;
-- repl
FUNCTION REP(line varchar) RETURN varchar IS
BEGIN
RETURN PRINT(EVAL(READ(line), ''));
END;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
BEGIN
plsql: step4 basics.
2016-04-05 01:29:14 -05:00
WHILE true LOOP
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
BEGIN
plsql: self-hosting. Long strings, refactor io.
2016-05-03 00:28:33 -05:00
line := io.readline('user> ', 0);
IF line = EMPTY_CLOB() THEN CONTINUE; END IF;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
IF line IS NOT NULL THEN
plsql: self-hosting. Long strings, refactor io.
2016-05-03 00:28:33 -05:00
io.writeline(REP(line));
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
END IF;
EXCEPTION WHEN OTHERS THEN
plsql: Fix readline/output async. Improve arg mode.
2016-05-03 22:29:07 -05:00
IF SQLCODE = -20001 THEN -- io read stream closed
io.close(1); -- close output stream
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
RETURN 0;
END IF;
plsql: self-hosting. Long strings, refactor io.
2016-05-03 00:28:33 -05:00
io.writeline('Error: ' || SQLERRM);
io.writeline(dbms_utility.format_error_backtrace);
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
END;
END LOOP;
END;
plsql: step2,3 basics.
2016-04-04 23:37:52 -05:00
plsql: refactor with memory pool. step5,7,8 basics. Ran into the same problem with macroexpand that prevented TCO in step5. Apparently objects as parameters cannot be re-assigning (or copied and re-assigned even) in a loop. When trying to do so, the server process crashes with a memory error. Move away from try to allocate and pass full structures around. Instead pass integer references into a memory pool of the objects (nested table of mal objects) and then pass this memory pool around by reference. Passing the memory pool around all over the place is ugly, but it allows TCO and macroexpand to work at all and seems to have better performance anyways.
2016-04-07 01:22:15 -05:00
END mal;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
/
plsql: step2,3 basics.
2016-04-04 23:37:52 -05:00
show errors;
plsql (Oracle PL/SQL): Copy plpgsql. Step0,1 basics.
2016-03-31 21:29:47 -05:00
quit;
Reference in New Issue Copy Permalink